Depletion of CD20 B cells increases active B cells…killing our idea…Maybe not.

To continue on with today’s subject
Are you interested in what happens in mice following CD20 depletion?
Read the paper below and I think it says we have got the “B memory cell idea” all wrong.
However, read the data (and the literatureL) and perhaps we can make some other conclusions. 

Suggesting mouse is not human… more ways than one


Häusler D, Häusser-Kinzel S, Feldmann L, Torke S, Lepennetier G, Bernard CCA, Zamvil SS, Brück W, Lehmann-Horn K, Weber MS.

Functional characterization of reappearing B cells after anti-CD20 treatment of CNS autoimmune disease.Proc Natl Acad Sci U S A. 2018 Sep 7. pii: 201810470.

The anti-CD20 antibody ocrelizumab, approved for treatment of multiple sclerosis, leads to rapid elimination of B cells from the blood. The extent of B cell depletion and kinetics of their recovery in different immune compartments is largely unknown. Here, we studied how anti-CD20 treatment influences B cells in bone marrow, blood, lymph nodes, and spleen in models of experimental autoimmune encephalomyelitis (EAE). Anti-CD20 reduced mature B cells in all compartments examined, although a subpopulation of antigen-experienced B cells persisted in splenic follicles. Upon treatment cessation, CD20+ B cells simultaneously repopulated in bone marrow and spleen before their reappearance in blood. In EAE induced by native myelin oligodendrocyte glycoprotein (MOG), a model in which B cells are activated, B cell recovery was characterized by expansion of mature, differentiated cells containing a high frequency of myelin-reactive B cells with restricted B cell receptor gene diversity. Those B cells served as efficient antigen-presenting cells (APCs) for activation of myelin-specific T cells. In MOG peptide-induced EAE, a purely T cell-mediated model that does not require B cells, in contrast, reconstituting B cells exhibited a naive phenotype without efficient APC capacity. Our results demonstrate that distinct subpopulations of B cells differ in their sensitivity to anti-CD20 treatment and suggest that differentiated B cells persisting in secondary lymphoid organs contribute to the recovering B cell pool.

Ocrelizumab inhibits active MS (round hole), therefore studies in EAE (Square peg) have the answer.
Perhaps I should say “let this lie” and not comment on this work
as it will do me no favours, but it is by opinion-leaders and it does seem to rubbish our ideas and the story we have been creating concerning memory B cells and MS.

Or Does it.

(However as they are not even cited L they may not be aware of them because they are not reading other peoples papers L because the literature review was not good enough or they are simply ignoring themL

We have made the suggestion that cells within the memory B cell compartment are the important cell types targeted by CD20- depletion in MS and other conditions and recently it has been shown in MS that the memory B cell population presents antigen to T cells in MS. Yet here they report inhibition of EAE with CD20 depletion despite an increase in CD27+ (memory cell marker in humans) cells. So it deserves comment…

So here is some of the data

CD20 depletion in mice can lead to a long-term (a few weeks) depletion of B cells in the blood and lymph glands and a quicker repopulation in bone marrow and the spleen. 
The authors state “we investigated three mechanistic questions that
cannot be addressed in humans”

  • First, we established that a
    fraction of mature B cells in the spleen is resistant to anti-CD20.
  • Second, we
    determined that, after cessation of treatment, splenic and bone-marrow B cells
    reconstitute in parallel, substantially preceding B cell reappearance in
  • Third, we observed that, in a model involving activated B cells,
    the post–anti-CD20 B cell pool contained an elevated frequency of
    differentiated, myelin-reactive B cells. Together, our findings reveal
    mechanisms by which pathogenic B cells may persist in anti-CD20 treatment”
So to me this rather says that B cell therapy should not work in
humans……..But it does! Or maybe they are saying CD20 depletion should not work in mice.
So perhaps All I should say is “Read the review above and perhaps one can see
that you can actually address these questions in humans” 

However, you know me too well. So off we go….

In this study they deplete with CD20-depleting antibody and then look at the cells repopulating and
they see an increase in CD27+ cells in both 
myelin oligodendrocyte glycoprotein  (MOG) protein and MOG peptide induced
But as MOG-induced disease is largely monophasic (a single attack) I
would say “So what! 
What does it tell us?” 

Not much, as disease activity will not resume in the
C57BL/6 mouse model no matter what happens to the cells, so we can’t determine the consequences of the changes.

The kinetics of B cell repopulation in the blood of humans is known.
There we see a stereotyped
repletion response with immature and then mature B cells showing repopulation,
probably from the bone marrow. There is a marked depletion of memory B cells in
virtually every study (see above). The mature (naïve) B cells is not the
important subset.
In addition, in this study they then looked at the B cell repertoire and
found there was a reduced diversity of repertoire, suggesting that the B cells
are initially repopulated from surviving B cells (I say….Surprise, Surprise?) but the
repertoire that survives is different. This is seen in MS (There are a number of studies on this but they were not cited in the discussion..Therefore so much for putting your work into the context of what is known).

In this mouse study, the B cells that return after MOG protein have more
antigen presenting function and they suggest that there are more autoreactive
cells. Surely th
is is opposite of what you would think you would want. Therefore what are these studies actually telling us…..I suppose that mouse studies are not going to give us a useful answer?
However does it tell us anything? In MS there is long-term benefit from a
short-term treatment. In mice there is no long-term benefit. So is the rapid repopulation by pathogenic cells the reason why there is no long-term treatment
effect in mice? 

Maybe, but I suspect not. In the C57BL/6 mice used, the attack causes nerve loss and this
does not recover. Hence you get the the flat-lining of the clinical disease. The data tells us
that the net result of a short term treatment of mice with CD20-depleting
antibody, whether it is MOG protein or MOG peptide means nothing in the long-term (See below).
However, first things first. This paper sets out to show “the unknown” like so
many other papers. 

The referee/editor buys into this and impact factor ten paper
here we come. 

However, I have to say that I hope the authors do not do their animal
experiments/pathology like their literature reviews

Simply read the paper above to see this.
You can indeed and do know, albeit it is only to some extent and not as
controlled as the animal experiments, what goes on in human lymph nodes/tonsils,
spleen, blood and bone marrow following rituximab.
For example the spleens are removed as a treatment aim in thrombocytic
purpura and this has occurred after rituximab, as have lymph node and bone
marrow biopsies. So the discussion of the paper is rather lacking to say the

“I’m just being constructive here”.
Next, also with a bit of reading it would become evident that human and
mouse B cell biology is not the same, perhaps because human B cell biology is
heavily influenced by Epstein Barr Virus, which does not infect rodents.
In humans memory cells can conveniently be detected by their expression
of CD27, but this appears not to be the case in mice, seemingly ignored by the
authors of this paper, who imply they are detecting antigen-activated (i.e. memory)
B cells by detecting CD27. They use CD138 to detect plasma cells, which I am not going to gripe about..
However, it has been reported by Xio et al. 2004 that It (CD27) is not a marker for somatically mutated B cells and is
present at very low frequency on memory B cells”
. Indeed memory B cells in mice are very complex and consist of multiple subtypes. Read Bergmann et al. 2013).

In this current study the inference is that the CD27 positive B cell
population go up and as disease is inhibited then the memory cell idea surely
falls down.  

However does it?
If you accept, which I don’t (see above), that CD27 is a type of memory cell
marker in mouse. 

The study shows that they are resistant to CD20 depletion . However, the work shows that certain B cell subsets are depleted and therefore
the percentage of the other stuff goes up, but without looking at the absolute
numbers of cells, we have no idea if this is a indeed a real increase in
numbers. They may be staying the same or going down. However, if we accept that the data
suggests that a population of B cells is less depleted by CD20 depletion.

Now let’s look at the therapy data. I am sorry to say it hardly passes the Smack
you in the Eye Test
 The control group in one experiment looks like the Test group
in the other experiment and
vice versa. The effects last a few days and suggests that the data is all within “experimental wobble” (biological variation),
meaning there is really no or very limited biological impact on the mice.

This is what our data showed and also found by others, indeed a poster  Boschert U et al by Merck (ISNI 2018 Brisbane) showed that CD20 antibody-B cell depletion had no impact on MOG-induced EAE in C57BL/6 mice. (How did the Bruton tyrosine kinase inhibitor work…it’s a macrophage blocker…but that’s another story)

In this study MOG protein induced disease, claimed to be B cell dependent, is inhibted by CD20 antibody depletion and MOG peptide induced disease claimed to be B cell independent is augmented by CD20 antibody depletion (see below)

Would you say that the animals are protected or do they get disease? 

The therapeutic effect in my eyes is marginal. Anti-CD20 depletion before immunization makes things worse in MOG35-55 group. OK I have seen this before, but let’s look at the data, the main difference is where the control group is. Also if a few animals (we don’t know as it is not in the main paper) don’t get disease the line is skewed making it very difficult to interpret.  

However, if the models (MOG peptide) are B cell independent, then why does anti-CD20 have a significant effect? 

This is interesting as most importantly the authors seem to have forgotten about their own data, published previously in  Molnarfi N, Schulze-Topphoff U, Weber MS, Patarroyo JC, Prod’homme T, Varrin-Doyer M, Shetty A, Linington C, Slavin AJ, Hidalgo J, Jenne DE, Wekerle H, Sobel RA, Bernard CC, Shlomchik MJ, Zamvil SSMHC class II-dependent B cell APC function is required for induction of CNS autoimmunity independent of myelin-specific antibodies. J Exp Med. 2013 210(13):2921-37.

Where they show and state “Thus, only rhMOG-induced (recombinant human = rh) EAE is considered B cell dependent (Lyons et al., 1999; Oliver et al., 2003; Marta et al., 2005). 

If it is only Human MOG that is B cell dependent, whereas mouse MOG and MOG35-55 are B cell independent. Then the effects seen here and previously (Weber et al. Anal Neurol  2010: 68:369) are surprising as an effect of B cell depletion is claimed in mouse MOG induced disease. If you look at the methods (Supplementary data) section they apparently use mouse MOG protein to induce EAE, so there should be no effect. 

So if the experiment is reproducible and the literature is correct, then results of the protein and peptide should have been the same, so maybe experimental wobble is indeed at work.….Oooops

Adding the two experiments together then suggests that there is no biological effect of depleting CD20 antibody in mice, which I think is probably the case.

Importantly, in this paper, the authors imply that CD20 depletion in humans is associated with increases in proliferating memory cells similar to this study in mouse, perhaps implying it is a good thing. However, if you read the references cited to justify this point it is clear that the human study implicates the occurrence of memory B cells to be associated with a worse prognosis and disease (re) activation, just as we have been banging on about.

So back to the paper. So we have CD20 depletion causing major depletion
of CD27  B cells and inhibition of relapse in humans, however if you actually look at the data there is no/minor impact (increase) on CD27
B cells and no/minor impact on clinical EAE.
So actually it all fits. The memory B cells may remain to
be an important subtype of cell. QED.
So we are still right. Is this crazy logic?
P.S. It doesn’t seem to bother the people putting square pegs into round

COI: I have none, but editor of paper is working in same place as an
author….Em…..Rant over 🙁

P.S. If this type of clinical data is indeed experimental wobble then it suggests that the conclusions of a large number of EAE papers that show similar types of clinical effects become meaningless and suggests that there will be little translational value of many EAE studies in predicting efficacy/biology in humans. 

But we know this, Don’t we?

P.S.S. If you are a scientist, this is surely a good Journal Club paper as it should teach you a lot on the need to read around papers to understand them and also teach you about data analysis (stats tests used in paper are all wrong as well:-0) and interpretation.

Maybe I have it all wrong

About the author



  • Thanks MD,

    It sounds like they should have read and implemented the ARRIVE guidelines, I wonder if the study had been blinded would those small differences have even occurred.

    • It may have been blinded but they just didn't report it…however it makes you more concerned about the validity of the results if it wasn't blinded.

      There are elements of ARRIVE that I think are nonsense but in principle the concept is a good thing and it is something I teach.

      I always say the result is the result and never pressure people to get result x or y (Sometimes you don't get what you expect), but this is a matter of scientific life or death in some labs and this is where it becomes critical to have best practice, because when you are doing subjective scoring as occurs in EAE, you have subliminal bias even when you don't think so..

      In general if I see such small differences as occurrred here it doesn't both me if there is a statistical difference as it doesn't get registered as something I should be really be interested in.

      I am instantly doubtful if there is any biological significance and believe unless proven otherwise that the data won't even be repeated in another lab let alone being important to humans. Sadly many people cannot see this and see such small differences as being of major importance. This is in part because it is the opinion leaders who are doing this.

      However it is another example of a over hype experimental result, of which there are so many examples in EAE, that it has turned my into a sad-old cynic. It is a shame that some neuros act on them and do a failed clinical study.

      Apply the "smack you in the eye test" and I think it is a good start. However knowledge from reading is a good insulator

  • "You can indeed and do know, albeit it is only to some extent and not as controlled as the animal experiments, what goes on in human lymph nodes/tonsils, spleen, blood and bone marrow following rituximab."

    Since Rituxan action its monitor in the blood could lymph nodes/tonsils,

    spleen, blood and bone marrow be shielding some resistant memory b cell?

    Aparently they say so

    A parallel
    observation was reported in patients with Sjögren’s syndrome,
    in whom persisting memory B cells could be detected in salivary
    glands even after 2 y of consecutive rituximab treatment (18). Along
    the same lines, anti-CD20 treatment of patients with rheumatoid
    arthritis enriched the relative abundance of memory B cells that
    coexpressed the proliferation marker Ki-67 (19), confirming that
    memory B cells can escape systemic anti-CD20–mediated B cell
    depletion, presumably in organs other than the blood.

    • Reference 18 was a review.

      The point they are making here is that memory B cells are not completely depleted by rituximab in lymphoid tissues. Yes depletion of B cells is less effective that the blood not just for memory cell. However they use this to support the idea that B cells escape depletion. Remember in humans this is associated with CD20 benefit.

      However, if you read the paper the point the authors of the paper in ref 19 are making are that those people who do not deplete memory B cells do worse. Therefore in reference 19 memory B cells = bad, in this study memory B cells = good.

      In figure 1B in spleen (This is easiest to work out) CD19 B cells are about 50% of the say 10,000 cells analysed, so that is say 5,000 B cells in the sample. Of the CD19 population about 5% are CD27+ so there about 250 cells.
      After rituximab there are about 5% CD19 cells so about 500 cells. Of these about 25% of these are CD27 so about 125 cells. So there is not an increase in CD27 cells as implied by the figure, so there is about 50% reduction in the CD27 cells not an increase. However it says the CD27 is less sensitive or they recover quickly.

  • Also they caution b cell regulatory role of myloid cells

    In parallel to this preclinical
    observation, we had reported earlier that, in patients with
    NMO and MS treated with anti-CD20, peripheral monocytes
    show signs of an enhanced activation status and proinflammatory
    differentiation (14), suggesting that B cells physiologically control
    the activity of myeloid cells and that this desirable B cell property
    is abolished by anti-CD20 treatment (15). The possible clinical
    relevance of this regulatory axis between B cells and cells of myeloid
    origin is highlighted by a recent case report in which a patient with
    NMO depleted of B and T cells by administration of alemtuzumab
    died after 20 mo of continuous deterioration, which was associated
    with a massive CNS infiltration of monocytes (16)

    Could you comment

    Nice work


    • The concept of regulatory B cells has been in existence for as long as I have been in science, indeed by boss talked about them and not t cells. It has been known for years that cyclophosphamide, which irradiates B cells because they have high endogenous turnover depletes a regulatory population if given before the antigen challenge, hence augmenting disease. Given after disease induction and it inhibits imune response. Therefore the balance is anti-inflammatory when used therapeutically. No doubt rituximab is similar. There is no reason to believe all it does it get rid of bad guys.

      As to the person on alemtuzumab, infection?
      Alemtuzumab is not recommended for NMO ( It is evidence that alemtuzumab augments antibody induced autoimmunity. NMO is just that.
      Antibodies will be bound by macrophages. This worsening happened over at least 8-10months

By MouseDoctor



Recent Posts

Recent Comments